This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PGA2310- volume change noise

Other Parts Discussed in Thread: PGA2310, NE5532, OPA2134, LME49720, PGA2310EVM, PGA4311

Hello,

I use PGA2310 for volume control in a system of balanced lines.

The sound is great, but I have a big issue when sending new volume command- I'm getting 'pops'- switching noise.

It is most noticed when the input is a clear sine wave, for example 400[HZ].

It seems that the zero crossing mechanism doesn't work well, or maybe I've missed something.

I'll appreciate any help.

Thank you in advance,

Yanki

  • How much loud is the pops noise? Normally a very small pop sound is produced but you can hear it only by placing your ear next to the speaker, with inputs shorted to ground and at big gain settings (say greater than -10dB). Regarding your test signal, yes i assume that it can incite pops that can be very clear, but with music signal there is no problem. A trick to avoid pops, is an appropriate code with 1dB steps up to -15dB and 0.5dB steps from -15dB and above. With the appropriate code, PGA2310 can mimics the volume setting of a traditional logarithmic audio potentiometer. Please note that PGA2310 has a flat frequency response up to 0dB (i.e. working as attenuator).
  • Hello Yanki,

    To enable the zero crossing detection pin 1 of the PGA2310 must be driven high. Can you please provide a schematic of your design so I can verify all the connections are correct?

    Thank you,

    Tim Claycomb

  • Hello Fotis & Tim,

    Thank you so much for your replies.

    I didn't reply sooner because I wanted to check few things:

    At the moment, I use it as an attenuator (up to 0[dB]). I can't accept any switch noise, since the device will be used by porffesional music producers.

    I'm using the PGA2310 according to the instructions in the datasheet. ZCEN is driven high. I tried to connect it to GND and got much worse results, so the connection is probably ok. Similar to the reference design in the datasheet.

    I read the datasheet very carefully with my board designer and we implement the instructions, including the separation of the digital and analog grounds, and connecting the input and output grounds at a single point. No lines are implemented under the PGA2310.

    I asked the board designer to provide the schematics. I'll post it soon.

    I've noticed few things:

    1. The higher the steps, the louder is the zipper noise. Steps of 0.5[dB] produce the minimal switching noise,

    2. When I use minimal steps (0.5dB)- I get louder noise when the delay between the steps is shorter.

    3. I don't get any switching noise when there is no input.

    Thank you both again,

    Yanki

  • Hello Yanki,

    Are the analog and digital grounds connected anywhere on the board? The datasheet specifies to keep the two grounds separate but connect the analog and digital grounds (not input/output grounds) at a single point.

    -Tim Claycomb

  • Hello all.

    my name is Isak, i'm working with Yanki on this project.

    the ground of the digital and analog are separate, they are linked only in one point the same as the datasheet mention.

    BTW, our processor is Arduino uno if its matter.

    searching about this matter over the and i found this amazing site by Mark Hennessy building his own preamp using the PGA2310 and this is what he found...

    "

    DC issues

    Upon completing the test board and ensuring it worked, the first thing I noticed was clicks and pops heard through the speakers while making volume adjustments. This was disappointing as the IC is meant to have zero-crossing detection to prevent such problems. Interestingly, these noises were much worse when the IC had gain rather than attenuation. So presumably, the zero-crossing detection wasn't working; to test this theory, I applied a low-level sinewave of around 30Hz (turning the bass to minimum on the workshop hi-fi) and found the effect was much better. Next I put a DC-coupled 'scope across the output of the input buffer, and varied the level of the input signal, clearly demonstrating that when the sinewave was large enough to cross zero, the clicks and pops went away. There was some -20mV of DC offset from the input buffer, caused by the input bias currents of the NE5532, and this was enough to cause these probles.

    At this point I began to wonder if I would have to include AC coupling between the input buffer and the PGA2310. This is obviously undesirable - the capacitor would have to be a large electrolytic to avoid the rising LF THD described above...

    The final answer was simple - use an OPA2134 for the input buffer. The FET-input stage results in less than a millivolt of offset, and the behaviour is much improved. As I mentioned earlier, Doug Self found some issues with this IC, but to be honest I think I can compromise here. Obviously, there are plenty of horrendously expensive ICs that might be better, so that's a subject for future research...

    The output NE5532 causes a similar DC offset on the output. However, this isn't a problem if the output is AC-coupled (using a large electrolytic bypassed by a non-electrolytic). That said, if I decide to use the input capacitor to define the system LF -3dB point, I can DC-couple the PGA2310 to the output buffer and this will reduce the offset, so there is a chance of omitting the output capacitors..."

    this is the website (page 10)

    www.markhennessy.co.uk/.../analogue_research.htm

    he also provided a scheme that shows exactly how he did the fix but even then i found no luck.

    he states that he didnt eliminate the click totally but got an satisfying results.

    we are using this encoder:

    www.mouser.co.il/.../

    as you will see in the datasheet there is no need to debounce this encoder ( i thought that the source of the click are from bouncing at first)

    another test we did was a code that command the PGA2310 going up and down automatically, no encoder involve , the clicks where there but not as loud with the encoder. 

    maybe it means something to you?

    Regards,

    Isak

     

  • Hello Isak,

    Thank you for providing the information above.

    Can you please provide the schematic of your design? Please know that this is a public forum so if you do not feel comfortable posting the schematic please let me know and you can email me directly.

    Thank you,

    Tim Claycomb

  • Hi Tim.

    Thank you very much for replying.

    Can you please give me your email?

    Cheers,

    Isak

  • Hi Isak

    Sorry for my first reply as is misinformating. I just did a check of my PGA2310 preamp and i can confirm you that is quite noiseless without any click or pop noise during volume up or volume down operation.

    From what i've read in your post, it seems that you have preciselly followed the guidelines provided in the PGA2310 datasheet regarding the necessary PCB layout. Am i right? Supposing that your PCB is of 2 layers: have you drawn two separate solid copper ground planes on the bottom layer under the PGA exactly as is described in datasheet?

    I don't think that your problem lies in the selection of the operational amplifiers used as input and output buffers. However in my project i use two LME49720 configured as voltage followers for this purpose. Another one difference is that i use an ALPS encoder equiped with detents, this one: MOUSER part: 688-EC11E18244A5. Your encoder is of optical type and is very expensive, however is your personal selection. The processor i use is a PIC16F1789. 

    Here is some code from my project that could offer you some ideas. Please note that i don't use the SPI interface of micro because is not available as i also use the I2S ports. For that reason i have obtained a custom code to provide controll on PGA2310 through the ports of micro RA.1=SCLK, RA.2=SDI, RA.3=CS. For code development i use the FlowCode graphical programming language. In the 1st picture is presented only a part of PGA_SDI macro as the rest steps to obtain the full gain byte are simillar.

    1) PGA_SDI:

    2) VOL_UP

    3) VOL_DOWN

    In VOL_UP and VOL_DOWN macros, you could see how the PGA2310 is encoded to simulate the operation of a logarithmic analog audio potentiometer. From actual tests the encoding of PGA to work in linear mode is inconvenient as the distance between the minimum level of -95.5dB (N=0) to the max +31.5dB (N=255) is so long and you must turn many times the rotary encoder. For that reason the logarithmic mode encoding is by far more convenient as with 3-4 full rotations of encoder you can move from the min to the max gain settling. For N=2 the lower gain limit is settled at -95dB. In my application i placed the max gain at +3dB and so the value of Nmax=198. Please note that for gain settings above 0dB the bandwidth (GBP) or the response in high frequencies of PGA2310 is reduced proportionally. The same applies for all operational amplifiers, that is why buffers (unity gain followers) present the best bandwidth. Where it is applicable don't exceed 0dB.          

  • Hello Isak,

    I have sent you a friend request. Please accept the friend request and I will provide you with my email.

    Thank you,

    Tim Claycomb

  • Hi Fotis Anagnostou

    thank you very much for your detailed reply.

    if i may ask, how exactly did you do the click test?

    did you run a sine wave or just a music?

    i can hear the clicks in sine wave, if you can run 400Hz, 500Hz, 10000Hz and then tell me if you still cant hear any click that would be cheering.

    i must add that i'm using it only as attanuator, not as a preamp, amp or any other amplifying thing.

    going out from my pro studio sound card (RME FF400) at 0dB (max volume in my sound card), and only attanuate with the PGA2310

    i'm going from -95dB up to 0dB (0-192), no more then that cause i dont need it to amplify at all.

    i did my own PCB, not professional printing, i etched my board, only one layer, so about the GND plane that you think i have, i dont.

    but i can almost be sure that the clicks problems are not cause of the GND plane.

    you can see the part of the PGA2310 in the pic below, nothing under the PGA2310, the GND's are across the board.

    if you want me to send you the full board design and scheme please give your email.

     

    thanks a lot for the help, really appreciate it!

    cheers,

    Isak  

  • Hello Isak,

    I have reviewed your schematic and do not see anything schematic related that may be causing your issue. As mentioned in my email to you, I recommend ordering a PGA2310EVM since not having a solid ground plane may be causing your issue.

    Can you verify, by probing the outputs of the PGA2310 with an oscilloscope (and be sure to DC couple the probes), that the gain is only switched after the output crosses zero twice? The PGA2310 datasheet mentions the zero point must be crossed twice or a time out of 16ms occurs for the gain to switch. If the zero point is not crossed twice, the gain will change 16ms later no matter what point the output is at. If the gain is changed after 16ms this may explain your issue.

    Thank you,

    Tim Claycomb

  • Hi guys.

    i found out the problem and the source of the noise.

    its the resolution of the PGA2310, the zero cross works, no GND problem at all.

    i did an audio clip going in and out of the sound card, i can see clearly the high jump steps especially in the high Vol adjustment.

    i did a test to proof my theory, i made the same audio i record by generating a sound with audio edit program, meaning, 500Hz sine wave, 1 cycle would be 2ms, 20ms delay between each step logarithmic.

    when playing the wave file i created the same audio clicks "zipper audio" as appeared in my studio monitors.

    when did another test it was more clear that its the resolution of the PGA2310

    i chain the right out of the PGA to the left input and from the left output to main out, i also change the scheme a bit.

    this way i increase the resolution by 4.

    when i did an audio record after the mods i did i got my clicks free PGA2310.

    i dint think my solution is a good solution cause i need balanced signal, meaning i will need 4 PGA's to make stereo balanced attenuator. 

    do you maybe have a better solution?

     

    regards,

    Isak

  • Hello Isak,

    Take a look at the PGA4311. The PGA4311 is a 4 channel PGA which would cut the required PGAs in half.

    Thank you,

    Tim Claycomb

  • Hi Tim,

    Thank you for your support.

    Can you suggest other solution for the low resolution issue?

    We wanted to order the evaluation board in order to eliminate board issues problems, but it seems to be a limitation of the chip.

    I can think of other solution, but all the 'work around' solutions don't feels like proper solutions, since it didn't meant to be so.

    We are very happy with the sound quality, but we can't accept the transition noise.

    Thank you again,

    Yanki

  • Hello Yanki,

    I apologize for a delay in my response. I was out of the office during the Holiday.

    Unfortunately I do not know of another device or solution that will decrease the step size when changing the gain. The PGA2310 has a step size of 0.5dB which to my knowledge is the smallest step size offered.

    Thank you,

    Tim Claycomb

  • Hi Timothy 

    how was the holidays? hope all went well :)

    well,  i was not sure that the board i design to the PGA2310 was ok so to eliminate this suspect i ordered the PGA2310 EVM.

    now i must say, it is final, the PGA clicks in every way, even in the EVM it self, this is very disappointing to know after wasted 8 month of research and a lot of internet hours to find the clicks problem which i believed was my fault when its actually the PGA2310 fault.

    sorry in advance if i sound rude but i must say that  this is defiantly not hi-end, not good for studio attenuater use, maybe for home theater preamp it will be good.

    i found 0.25dB step IC, not TI's, after simulate the audio in software i will go with this one.

    i thank you for the willing to help.

    regards,

    Isak